The First Phylogenetic Analysis of Palpigradi (Arachnida)—The Most Enigmatic Arthropod Order Citation Gonzalo, Gilbert, McIntyre Erin, Christian Erhard, Espinasa Luis, Ferreira Rodrigo L., Francke Óscar F., Harvey Mark S., Isaia Marco, Kováč Ĺubomír, McCutchen Lynn, Souza Maysa F. V. R., and Zagmajster Maja. 2014. "The First Phylogenetic Analysis of Palpigradi (Arachnida) – The Most Enigmatic Arthropod Order." Invertebrate Systematics 28: 350–360. doi: 10.1071/IS13057 Published Version doi:10.1071/IS13057 Permanent link http://nrs.harvard.edu/urn-3:HUL.InstRepos:12313557 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Open Access Policy Articles, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#OAP Share Your Story The Harvard community has made this article openly available. Please share how this access benefits you. Submit a story . Accessibility 1 The first phylogenetic analysis of Palpigradi (Arachnida)—the most 2 enigmatic arthropod order 3 Gonzalo GiribetA,K, Erin McIntyreA, Erhard ChristianB, Luis EspinasaC, Rodrigo L. FerreiraD, Óscar F. 4 FranckeE, Mark S. HarveyF, Marco IsaiaG, Ľubomīr KováčH, Lynn McCutchenI, Maysa F. V. R. 5 SouzaD and Maja ZagmajsterJ 6 7 AMuseum of Comparative Zoology, Department of Organismic and Evolutionary Biology, Harvard 8 University, 26 Oxford Street, Cambridge, MA 02138, USA. 9 BInstitut für Zoologie, Universität für Bodenkultur, Gregor-‐Mendel-‐Straße 33, 1180 Wien, Austria. 10 CSchool of Science, Marist College, 3399 North Road, Poughkeepsie, New York, USA. 11 D Centro de Estudos em Biologia Subterrânea, Departamento de Biologia, Universidade Federal 12 de Lavras, Lavras, MG. CEP 37200-‐000, Brazil. 13 EColección Nacional de Arácnidos, Instituto de Biologia, UNAM, Apartado Postal 70-‐153, C. P. 14 04510, Mexico, D. F., Mexico. 15 FDepartment of Terrestrial Zoology, Western Australian Museum, Locked Bag 49, Welshpool DC, 16 WA 6986, Australia. 17 GDipartimento di Scienze della Vita e Biologia dei Sistemi, Università di Torino, Via Accademia 18 Albertina 13, 10123 Torino, Italy. 19 HDepartment of Zoology, Institute of Biology and Ecology, Faculty of Science, P. J. Šafárik 20 University, Moyzesova 11, 040 01 Košice, Slovakia. 21 IDepartment of Biology, Kilgore College, 1100 Broadway, Kilgore, TX 75662, USA. 22 JSubBioLab, Department of Biology, Biotechnical Faculty, University of Ljubljana, Večna pot 111, 23 SI-‐1000 Ljubljana, Slovenia. 24 KCorresponding author. Email: [email protected] 1 25 Abstract. Palpigradi are a poorly understood group of delicate arachnids, often found in 26 caves or other subterranean habitats. Concomitantly, they have been neglected from a 27 phylogenetic point of view. Here we present the first molecular phylogeny of palpigrades based 28 on specimens collected in different subterranean habitats, both endogean (soil) and hypogean 29 (caves), from Australia, Africa, Europe, South America and North America. Analyses of two 30 nuclear ribosomal genes and COI under an array of methods and homology schemes found 31 monophyly of Palpigradi, Eukoeneniidae, and a division of Eukoeneniidae into four main clades, 32 three of which include samples from multiple continents. This supports either ancient vicariance 33 or long-‐range dispersal, two alternatives we cannot distinguish with the data at hand. In 34 addition, we show that our results are robust to homology scheme and analytical method, 35 encouraging further use of the markers employed in this study to continue drawing a broader 36 picture of palpigrade relationships. 37 38 39 Additional keywords: Arachnida, micro-‐whip scorpions, palpigrades, speleobiology, 40 biogeography. 2 41 Introduction 42 The arachnid order Palpigradi (micro-‐whip scorpions or palpigrades) is one of the smallest, 43 rarest and most neglected groups of terrestrial arthropods, and one of the last arachnid orders 44 to be discovered—it was first reported only in 1885 (Grassi and Calandruccio 1885). The first 45 photographs of living palpigrades did not appear published until the first decade of the 21st 46 century (Kováč et al. 2002; Beccaloni 2009). Additionally, only a handful of DNA sequence data 47 are available in GenBank; with only 64 sequences, 56 are for Prokoenenia wheeleri (Rucker, 48 1901), a species that was part of a multi-‐gene phylogeny of arthropods (Regier et al. 2010), 49 while the remaining eight sequences are unidentified specimens from three studies on 50 chelicerate phylogenetics (Giribet et al. 2002; Pepato et al. 2010; Arabi et al. 2012). Contrary to 51 this, one can find more DNA sequences for other small arachnid orders in GenBank: 105 for 52 Uropygi, 200 for Schizomida, 200 for Ricinulei, 251 for Amblypygi, and 502 for Pseudoscorpiones, 53 [checked on October 25th, 2013]. In addition, there are only 2 sequences available in the 54 Barcode of Life website (http://www.barcodinglife.org). 55 Palpigrades are delicate animals that walk sensing the substrate with what seems a nervous 56 behaviour of the first pair of walking legs, and use their unmodified palps for walking, unlike all 57 other arachnids (Fig. 1). While moving, most palpigrades keep the flagellum upward, moving it 58 laterally. Accordingly, it is possible that the uplifted flagellum is associated with perception of 59 the environment (Ferreira and Souza 2012). These small, depigmented and highly translucent 60 arachnids range in size from 0.65 mm in Eukoenenia grassii (Hansen, 1901) to 2.4 mm in the 61 “giant” E. draco (Peyerimhoff, 1906) from caves on the island of Majorca (Mayoral and Barranco 62 2013). Eukoenenia spelaea (Peyerimhoff, 1902) from Slovakia has recently been reported to 63 feed on heterotrophic Cyanobacteria (Smrž et al. 2013). The mode of sperm transfer in these 64 arachnids remains unknown. 65 The living members of the order are currently divided in two families, Eukoeneniidae 66 Petrunkevitch, 1955, with 4 genera and 85 named species, and Prokoeneniidae Condé, 1996, 67 with 2 genera and 7 named species (Harvey 2002; Prendini 2011; Souza and Ferreira 2013). 68 Eukoeneniidae includes the genera Allokoenenia Silvestri, 1913 (1 sp. from West Africa), 69 Eukoenenia Börner, 1901 (71 spp., on all continents under tropical and subtropical climate; in 70 temperate regions predominantly in caves), Koeneniodes Silvestri, 1913 (8 Palaeotropical spp.) 71 and Leptokoenenia Condé, 1965 (5 spp. in the Afrotropical, Neotropical and Palearctic regions). 3 72 Prokoeneniidae includes the genera Prokoenenia Börner, 1901 (6 spp. in the Nearctic, 73 Neotropical and Oriental regions) and Triadokoenenia Condé, 1991 (1 sp. from Madagascar). 74 Further unnamed new species are known to us from various parts of the world. 75 The position of Palpigradi among the arachnid orders remains highly debated. The largest set of 76 data analysed to date places them as the sister group to Acariformes mites in a basal position 77 within arachnids, although without support (Regier et al. 2010). The most recent morphological 78 cladistic analysis of arachnid relationships leaves them mostly unresolved among the clades 79 Stomothecata, Haplocnemata, Pantetrapulmonata, and Acaromorpha (Shultz 2007). Earlier 80 studies combining morphology and a small set of molecular data placed Palpigradi as the sister 81 group of Ricinulei + Tetrapulmonata or as sister to Pycnogonida when fossils were considered, 82 although again, without significant clade support (Giribet et al. 2002); as sister to a clade 83 including Acari and Solifugae, based on the same two markers used in earlier studies (Pepato et 84 al. 2010); or in an unresolved position within arachnids (Arabi et al. 2012). Even less is known 85 about the internal relationships of the group, since no published study—molecular or 86 morphological—has yet incorporated information for more than one palpigrade species, and 87 only one unpublished masters thesis has explored palpigrade relationships cladistically, using 88 morphology (Montaño Moreno 2008). 89 To bridge this important gap in the knowledge of this arachnid order, although acknowledging 90 the difficulties in sampling and identification of these elusive animals, we obtained samples for 91 as many species of palpigrades as possible and from as many localities as possible with the aim 92 to obtain molecular DNA sequence data to generate a first hypothesis of internal palpigrade 93 relationships. 94 95 Materials and Methods 96 Taxon sampling 97 Palpigrades are difficult to obtain and identify, and success of field sampling differed among 98 regions included in the study. In Western Australia, many samples were collected indirectly in 99 caves and bore holes. In Brazil and Europe, they can be abundant in caves, where fresh 100 specimens have recently become available for inclusion in molecular studies. Additional samples 4 101 were from soil samples in Australia, Italy and the USA. In addition to fresh material collected for 102 this study, older specimens were used, especially from the diverse cave systems in Brazil, where 103 several new species have been recently described (Souza and Ferreira 2010; Ferreira et al. 2011; 104 Souza and Ferreira 2011a; Souza and Ferreira 2011b; Souza and Ferreira 2012a; Souza and 105 Ferreira 2012b). While a recently collected specimen of Eukoenenia ferratilis Souza & Ferreira, 106 2011 amplified well for some of the studied markers, none of the six specimens of Allokoenenia 107 spp. and the two specimens of Leptokoenenia sp. collected from the caves yielded workable 108 DNA. We also obtained a relatively large collection of specimens from the Western Australian 109 bore holes from Barrow Island and the Pilbara, but these were collected from litter traps and 110 many specimens did not amplify or only yielded some amplicons. Some of these specimens are 111 probably related to the Western Australian endemic E. guzikae Barranco & Harvey, 2008, but 112 unrelated to the more widespread species E. mirabilis (Grassi & Calandruccio, 1885), also found 113 in Western Australia (Harvey et al. 2006; Barranco and Harvey 2008). A single specimen of 114 Prokoenenia wheeleri was obtained from the Austin area (Texas, USA), but amplified well for all 115 fragments attempted. In addition, we obtained samples of Eukoenenia mirabilis from Italy 116 (Christian et al. 2010) and Australia (Harvey et al. 2006), E. spelaea (Peyerimhoff, 1902) from 117 multiple localities in Slovenia and Slovakia (Kováč et al. 2002; Zagmajster and Kováč 2006; Král et 118 al. 2008). Italian samples also include E. bonadonai Condé, 1979 and E. strinatii Condé, 1977, 119 collected in caves. We also included specimens from multiple localities from the hanseni-‐ 120 chilanga group of Eukoenenia from Mexico and the USA (Montaño-‐Moreno 2012). Additional 121 specimens come from Mexican caves and South Africa. Details on collecting localities are 122 available in Table 1 and in MCZBASE (http://mczbase.mcz.harvard.edu/SpecimenSearch.cfm). 123 Vouchers or additional specimens are deposited in the Museum of Comparative Zoology, 124 Harvard University (MCZ), and in the Western Australian Museum (WAM). 125 We included three species available in GenBank, one from South Africa sequenced by 126 Giribet et al. (2002), one from Brazil from Pepato et al. (2010), and one of unknown origin 127 published by Arabi et al. (2012). Here we added sequences from an additional South African 128 specimen from the same collection of that from Giribet et al. (2002), and a specimen of E. 129 ferratilis from Brazil, which was identical to the specimen reported by Pepato et al. (2010) as 130 Eukoenenia sp., and to which we refer to as E. cf. ferratilis in the present study. Outgroup taxa 131 were selected from GenBank (Table 2), mostly from previous studies on arthropod or arachnid 132 phylogeny using nuclear ribosomal genes (Giribet et al. 2002; Mallatt and Giribet 2006). 5 133 134 Molecular methods 135 Although we attempted to amplify and sequence five molecular markers typically used in other 136 analyses of arachnid systematics (e.g., Dimitrov et al. 2012; Giribet et al. 2012), the 137 mitochondrial 16S rRNA gene only amplified for Prokoenenia wheeleri and the nuclear protein-‐ 138 encoding gene histone H3, although amplified for several samples, did not produce clean reads. 139 We thus restricted our study to the two broadly available nuclear ribosomal genes, the 140 complete 18S rRNA and ca. 2.2 Kb of 28S rRNA, and the mitochondrial protein-‐encoding 141 cytochrome c oxidase subunit I (COI hereafter) (as in Murienne et al. 2008), although the latter 142 gene only amplified for about a third of the specimens (Table 1). For two of the bore-‐hole 143 Western Australian specimens, poorly preserved, only the middle amplicon of 28S rRNA worked. 144 Total DNA was extracted from whole specimens or from the opisthosomal region using 145 Qiagen’s DNEasy® tissue kit (Valencia, CA, USA). Although we were aiming to preserve the 146 digested carcass as a morphological voucher, it was completely digested and not recoverable. 147 Purified genomic DNA was used as a template for Polymerase chain reactions (PCR) 148 amplification. PCR, visualization by agarose gel electrophoresis, and direct sequencing were 149 conducted for most specimens as described in earlier work, e.g., Edgecombe and Giribet (2009). 150 Chromatograms obtained from the automatic sequencer were read and sequences assembled 151 using the sequence editing software Sequencher™ (Gene Codes Corporation, Ann Arbor, MI, 152 USA). Sequence data were edited in MacGDE (Linton 2005). The three genes were analysed as 153 follows: 154 18S rRNA: This marker was amplified in three amplicons (a, b, c), as in previous studies 155 (Edgecombe and Giribet 2009; Giribet et al. 2010; Giribet et al. 2012). In the present study we 156 include 27 palpigrade specimens plus 8 outgroups, for a total of 1760-‐1771 bp per complete 157 sequence (up to 1805 bp for one of the outgroups). From the 27 palpigrade sequences all but 158 three were complete; E. spelaea is missing fragment a and the sample of Eukoenenia from South 159 Africa (DNA100456.2) is missing fragment b. For the direct optimization analyses the three 160 amplicons were treated as a single input file, containing 23 sequences, and divided into six 161 fragments. The three amplicons were concatenated for the static alignment analyses. 6 162 28S rRNA: This nuclear gene was amplified in three amplicons (a, b, c), as described in 163 Giribet and Shear (2010). The data set includes 29 palpigrade specimens plus 8 outgroups, for a 164 total of 2,150 to 2,204 bp, with some length variation among species. These three fragments 165 correspond to primer pairs 28S rd1a—28D rd4b, 28Sa—28S rd5b, and 28S rd4.8a—28S rd7b1. 166 Some of the published sequences were amplified with a shorter fragment b, generated with 167 primers 28Sa—28Sb (Whiting et al. 1997), and therefore fragment b was divided into fragments 168 b1 and b2 to accommodate these two amplicons. Fragment a was available for 22 palpigrades 169 and divided into three fragments, fragment b for 29 palpigrades and three fragments, and 170 fragment c for 25 palpigrades and analysed as a single fragment. These were treated as three 171 different amplicons for the dynamic homology analyses, but aligned together for the static 172 homology approaches. 173 COI: This widely used mitochondrial marker amplified for ten palpigrade terminals in a 174 single amplicon using primers LCO—HCO, showing no length variation (654 bp analysed), plus 175 one was available in GenBank. COI did not amplify for many individuals, perhaps due to major 176 changes in this marker, as evidenced by the deletion of one amino acid with respect to the 177 outgroups. Five outgroup sequences were obtained from GenBank, but these were 3 bp longer 178 in all cases except for the pseudoscorpion. It was analysed as a single fragment; not pre-‐aligned 179 due to the length difference with some outgroups. 180 181 Phylogenetic analyses 182 Parsimony analyses were based on a direct optimization (DO) approach (Wheeler 1996) using 183 POY v. 5.0 (Varón et al. 2012). Tree searches were performed using the timed search function in 184 POY, i.e., multiple cycles of (a) building Wagner trees, (b) subtree pruning and regrafting (SPR), 185 and (c) tree bisection and reconnection (TBR), (d) ratcheting (Nixon 1999), and (e) tree-‐fusing 186 (Goloboff 1999, 2002) [command: search (max_time:00:01:00, min_time:00:00:10, 187 ]. For the individual partitions, timed searches of 1 hour were run on hits:20, memory:gb:2) 188 4 processors under six parameter sets, as in Giribet et al. (2012) (see Table 3). For the combined 189 analysis of the three markers we started with the same search strategy, giving the 28S rRNA 190 trees as input—as these contained all the taxa in the combined data set—, and the resulting 191 trees were given as input for a second round of analyses (sensitivity analysis tree fusing; SATF), 7 192 as described by Giribet (2007), and continued until the tree lengths stabilised (Giribet et al. 193 2012). The optimal parameter set was estimated using the modified ILD metrics (Wheeler W 194 1995; Sharma et al. 2011), as a proxy for the parameter set that minimizes overall incongruence 195 among data partitions (Table 4). Nodal support for the optimal parameter set was estimated via 196 jackknifing (250 replicates) with a probability of deletion of e-‐1 (Farris et al. 1996) using 197 , as discussed in earlier work (Giribet et al. 2012). auto_sequence_partition 198 Maximum likelihood (ML) analyses were conducted on static multiple sequence 199 alignments (MSA) inferred in MUSCLE v. 3.6 (Edgar 2004) through the EMBL-‐EBI server 200 (http://www.ebi.ac.uk/Tools/msa/muscle/). We also used an implied alignment (IA) generated 201 in POY (Wheeler 2003; Giribet 2005) for subsequent analyses based on static alignments, as 202 recently explored by Giribet and Edgecombe (2013b) for a centipede data set. The MUSCLE 203 alignments were conducted for each gene independently. The IA and MSA therefore were based 204 on the same data (see length for each gene in Table 5). In order to evaluate the impact of the 205 hypervariable regions in the data set, MSAs and IAs were subsequently trimmed with Gblocks v. 206 0.91b (Castresana 2000; Talavera and Castresana 2007) to cull positions of ambiguous homology 207 (see length for each trimmed gene in Table 5). In the case of 28S, fragments a and bc were 208 Gblocked separately, due to the larger proportion of missing data in the a fragment, which 209 otherwise would be deleted from the final 28S alignment. These data sets are thus based on 210 different data from their original sources and from each other, but the remaining data use the 211 same homology scheme as the source. Data sets were concatenated with SequenceMatrix 212 (Vaidya et al. 2011). 213 Maximum likelihood analyses were conducted using RAxML ver. 7.2.7 (Stamatakis et al. 214 2008b) in the CIPRES server (Miller et al. 2010). For the searches, a unique General Time 215 Reversible (GTR) model of sequence evolution with corrections for a discrete gamma 216 distribution (GTR + Γ) was specified for each data partition, and 100 independent searches were 217 conducted. Nodal support was estimated via the rapid bootstrap algorithm (1000 replicates) 218 using the GTR-‐CAT model (Stamatakis et al. 2008a). Bootstrap resampling frequencies were 219 thereafter mapped onto the optimal tree from the independent searches. 220 In total we analysed five data sets accounting for different optimality criteria, homology 221 schemes, and/or amount of data, as follows: 8 222 (cid:1) Analysis 1. Direct optimization/dynamic homology under parsimony (full sensitivity 223 analysis of 6 parameter sets) analysed in POY 224 (cid:1) Analysis 2. Static homology from the implied alignment for the optimal parameter 225 set under ML (analysed in RAxML) 226 (cid:1) Analysis 3. Static homology from the implied alignment for the optimal parameter 227 set trimmed with Gblocks under ML (analysed in RAxML) 228 (cid:1) Analysis 4. Static homology based on MUSCLE multiple sequence alignment 229 (analysed in RAxML) 230 (cid:1) Analysis 5. Static homology based on MUSCLE/Gblocks (analysed in RAxML) 231 232 Results and Discussion 233 All phylogenetic analyses yielded very similar results with respect to the ingroup relationships, 234 while the outgroup relationships were incongruent from analysis to analysis and unsupported 235 for the most part (Figs. 2 and 3). The latter was expected given the small amount of data and 236 outgroup taxa and the poor resolution in deep arachnid relationships in other studies (e.g., 237 Wheeler and Hayashi 1998; Giribet et al. 2002; Pepato et al. 2010; Regier et al. 2010). The 238 optimal parameter set under parsimony direct optimization was 3211 (where indel opening 239 costs 3, indel extension 1, transversions cost 2 and transitions cost 1; ILD = 0.00913), with a W 240 cost of 10,408 weighted steps (Fig. 2). Nearly all examined parameter sets concurred on the 241 topology of the optimal parameter set, with the exception of Eukoenenia spelaea IZ-‐19346 from 242 Slovenia, and the resolution of one of the Eukoenenia clades (see below). Likewise, the analyses 243 of the four data sets analysed under maximum likelihood were nearly identical, except for some 244 of the shallowest relationships. One of these trees, the one for the multiple sequence alignment 245 trimmed with Gblocks—the one that could be potentially the most different from the POY 246 analysis—is presented in Fig. 3, and it is virtually identical to the direct optimization tree. From 247 the 10 nodes depicted in Fig. 2 summarizing the six direct optimization and the four maximum 248 likelihood analyses, 5 were recovered in all analyses. Support values for these five nodes is high 249 for most analyses (jackknife values are lower by definition), with the exception of clades III and 250 IV in the DO analysis. Basically, nearly all analyses concur on the overall topology of the 251 palpigrade tree. 9
Description: